China is putting small and medium-sized enterprises at the center of its next industrial AI push. People’s Daily reported on September 5 that the Ministry of Industry and Information Technology had issued the Artificial Intelligence SME Entrepreneurship Support Plan for 2026 to 2028, setting targets for at least 10,000 SMEs designated as technologically innovative and more than 2,000 specialized Little Giant companies.
The program is notable because it makes AI entrepreneurship a structured policy objective rather than leaving it solely to large cloud companies and national champions. The plan calls for 10 high-standard technology-enterprise incubators, 10 national SME public-service demonstration platforms, and 10 national-level SME characteristic industrial clusters focused on AI. Those numbers give local governments and industry organizations concrete targets for the next three years.
The source does not say that all 10,000 firms will be newly created. The objective covers enterprises given an innovation-oriented SME designation, including companies at different stages of development. Likewise, the Little Giant target refers to China’s program for specialized, sophisticated small enterprises with distinctive technologies. The policy is therefore about raising the capabilities and specialization of a broad supplier base, not merely adding startup registrations.
The Plan Connects AI to China’s Existing SME System
MIIT’s approach builds on China’s established industrial-policy infrastructure. Incubators can provide workspace, mentoring, financing connections, and technical support. National public-service demonstration platforms can link smaller firms to testing, standards, computing resources, or procurement opportunities. Industrial clusters can concentrate companies, universities, component suppliers, and customers in one location.
The AI plan uses all three tools. A small robotics software firm, for example, may need access to hardware test sites and manufacturing partners. A model developer may need computing capacity, data governance advice, and enterprise customers. A company making industry-specific AI tools may need a path into a local supply chain. The document’s targets suggest Beijing wants local ecosystems that can solve these practical problems rather than a program centered only on research grants.
People’s Daily reported that China announced the plan around the 32nd APEC SME Ministerial Meeting in Guangzhou. The country also launched an Asia-Pacific AI Empowerment Cooperation Center for SMEs and a dedicated SME cooperation zone in Shenzhen. These initiatives extend the story beyond domestic entrepreneurship by linking it to regional training and technology cooperation.
China said it would provide 5,000 AI training and seminar opportunities for developing countries over the next five years, according to the report. The article also said AI application cooperation centers with ASEAN partners already operate in Laos, Malaysia, and Vietnam, supporting cooperation in smart agriculture and data industries. The training commitment is an announced government initiative, not a measure of completed programs.
The policy arrives after earlier efforts to enlarge China’s AI service ecosystem. EastFrontier reported on September 1 that MIIT set a 3,000-provider goal for China’s AI services. The new SME plan adds a different layer: it targets the smaller companies that can adapt general models, robotics tools, chips, and cloud services to particular industries.
Little Giants Could Become the Applied AI Supply Chain
The Little Giant designation matters because China’s AI economy needs specialized components as well as foundation models. Large companies can train models and operate cloud platforms, but smaller firms often supply sensors, industrial software, domain data, testing tools, specialized chips, and system-integration services. An AI application may depend on a network of companies that are not household names.
The People’s Daily report cites OECD data presented at the ministerial meeting saying SMEs contribute about 70% of China’s technological innovation and 60% of its gross domestic product. Those figures should be attributed to the OECD data, not treated as a new MIIT calculation. Still, they explain why policy makers are putting SME development alongside headline investment in computing infrastructure and frontier models.
The plan’s emphasis on 10 industrial clusters also points to a geographic strategy. A cluster could be organized around a regional manufacturing base, a university center, a robotics zone, or a cloud-computing hub. Its value will depend on whether it can connect firms to customers and provide services that individual startups cannot afford by themselves.
The details of funding, computing subsidies, tax treatment, and selection criteria are not set out in the People’s Daily article. Those gaps matter. A target for 10,000 companies is easier to announce than to turn into viable businesses. Smaller firms will need revenue-generating customers, reliable access to chips and cloud services, and the ability to protect their intellectual property in a crowded market.
China’s recent startup data illustrate the scale of the opportunity and the competition. EastFrontier reported in August that China’s generative-AI startups grew 28% in the first half as 55,000 new firms entered the market. A larger startup population can create more experimentation, but it also makes it harder for individual firms to find differentiated products and sustain margins.
Success Will Depend on Commercial Adoption, Not Targets Alone
The practical test of the 2026–2028 plan will be whether the targeted SMEs can turn AI demonstrations into long-term products. A company that builds an agent for one factory or a vision tool for one logistics provider may need to adapt its software for other customers. That requires domain knowledge, implementation teams, customer trust, and sometimes integration with older industrial systems.
The Asia-Pacific cooperation elements may add another route to scale. Smart agriculture, trade, and data-industry projects in ASEAN countries can become reference cases for Chinese SMEs that want to export applied AI. The report does not identify participating companies or contract values, so it is too early to treat the centers in Laos, Malaysia, and Vietnam as proof of a large export market.
The plan also has a workforce dimension. Training programs can improve familiarity with AI, but a seminar is not the same as building an engineering team or operating a secure enterprise system. Smaller firms may struggle to recruit technical staff when large technology companies can offer higher salaries and broader computing resources.
For policy makers, the value of the SME approach is resilience. A diverse base of companies can create more paths from research to commercial application and reduce dependence on a small number of platforms. For investors, it creates a larger funnel of potential partners and acquisition targets, but also a more crowded field where many companies may not survive.
MIIT’s targets are ambitious but concrete. By 2028, the government wants thousands of innovative AI SMEs, thousands of specialized Little Giants, and a national network of incubators, platforms, and clusters. The plan’s success will be measured less by the number of labels awarded than by whether those companies develop tools that customers in manufacturing, agriculture, services, and other sectors keep using after the initial policy support ends.